Journal of Theoretical and Applied Information Technology

©2005 - 2008 JATIT. All rights reserved.

www.jatit.org

HOTTEST SPOT AND LIFE EVALUATION OF POWER

TRANSFORMER DESIGN USING FINITE ELEMENT
METHOD
1
Ankireddypalli S. Reddy, 2 Dr M. Vijaykumar
1
Research Scholar, JNT University, Hyderabad, India.
2
Prof. & Head, Electrical Engineering Department, JNTU CEA, Anantpur, India
E-mail: svas_a@rediffmail.com, mvk_2004@rediffmail.com

ABSTRACT

Power transformers represent the largest portion of capital investment in transmission and distribution
substations. In addition, power transformer outages have a considerable economic impact on the operation
of an electrical network. One of the most important parameters’ governing a transformer’s life expectancy
is the hot-spot temperature value. Due to under estimation of stray losses in power transformer hot spots
are formed. So the stray loss evaluation is an essential aspect to calculate hot spot temperature. The stray
losses in transformer is caused by the time variable leakage flux which induces emf’s & circulates eddy
currents in the winding conductors and other conducting parts of transformer like tank wall, core, clamps
etc.. Evaluation of stray losses can be done more accurately by Finite Element Method (FEM). This paper
presents the new and more accurate temperature calculation methods taking into account the above
mentioned findings to estimate life of transformer form hot spot temperature.
Keywords: Finite element method – Hotspot - Power transformer - Stray losses - Thermal model

1. INTRODUCTION transformer loading (overshoot time dependent

POWER transformers represent the largest portion
of capital investment in transmission and
distribution substations. In addition, power
transformer outages have a considerable economic
impact on the operation of an electrical network.
One of the most important parameters’ governing
a transformer’s life expectancy is the hot-spot
temperature value [1]. Due to under estimation of
stray losses in power transformer [2, 3] hot spots
are formed. So the stray loss evaluation is an
essential aspect to calculate hot spot temperature.
the stray losses in transformer is caused by the
time variable leakage flux which induces emf’s &
circulates eddy currents in the winding conductors
and other conducting parts of transformer like tank
wall, core, clamps etc.. Evaluation of stray losses
can be done more accurately by FEM.
The classical approach has been to consider the
hot-spot temperature as the sum of the ambient
temperature, the top-oil temperature rise in tank,
and the hot-spot-to-top-oil (in tank) temperature
gradient When fiber optic probes were taken into
use to record local hot-spots in windings and oil
ducts, it is noticed that the hot-spot temperature
rise over top-oil temperature due to load changes is
a function depending on time as well as the
function). It has also been noticed that the top-oil
temperature time constant is shorter than the time
constant suggested by the present IEC loading
guide, especially in cases where the oil is guided
through the windings in a zigzag pattern for the
ONAN and ONAF cooling modes [9]. This results
in winding hottest spot temperatures higher than
those predicted by the loading guides during
transient states after the load current increases,
before the corresponding steady states have been
reached [4 -8].
This paper presents new and more accurate
temperature calculation methods taking into
account the losses including the chief cause for hot
spot i.e. stray loss [10 – 12], which is evaluated by
using finite element method (FEM). The models
are based on heat transfer theory, application of the
lumped capacitance method, the thermal-electrical
analogy and a new definition of nonlinear thermal
resistances at different locations within a power
transformer. The methods presented in this paper
take into account oil viscosity changes and loss
variation with temperature. The changes in
transformer time constants due to changes in the
oil viscosity are also accounted for in the thermal
models. In addition, the proposed equations are
used to estimate the equivalent thermal

238
 Journal of Theoretical and Applied Information Technology

©2005 - 2008 JATIT. All rights reserved.

www.jatit.org

capacitances of the transformer oil for different ⎡ A1 ⎤

(t + Δt )
transformer designs and winding-oil circulations.
where, A(t + Δt )) = ⎢⎢ A2⎥⎥ and
Here, an important and crucial part of transformer ⎢⎣ A3⎥⎦
losses i.e. stray loss, which decides the hot spot,
evaluation is explained in section II, where as ⎡ A1 + 0.5 A2 + 0.5 A3⎤
(t )
⎡1⎤
section III is to focus about the thermal modeling σD ⎢ ⎥ Js (t + Δt ) ⎢ ⎥
Q= ⎢0.5 A1 + A2 + 0.5 A3⎥ + ⎢1⎥
of transformer. The results are discussed in section 12Δt 6
IV and section V gives the conclusion. ⎢⎣0.5 A1 + 0.5 A2 + A3⎥⎦ ⎢⎣1⎥⎦

II. STRAY LOSS EVALUATION USING FEM is the right hand side vector, containing source
terms resulting from applied currents, and induced
A. Stray loss / Eddy current evaluation using
currents at the previous time step. In the left hand
FEM
side of the equation we have,
Equations defining magneto dynamic fields are
∂B ⎡ q1q1 + r1r1 q1q2 + r1r 2 q1q3 + r1r3 ⎤
∇ΧH = J ; ∇ΧE = − and ∇.B = 0 γ ⎢
∂t S =− q1q2 + r1r 2 q2q2 + r 2r 2 q2q3 + r 2r3⎥
For magneto-dynamic fields, the flux density is 2D ⎢ ⎥
⎣⎢q1q3 + r1r3 q2q3 + r 2r3 q3q3 + r3r3⎦⎥
time variant. The problem then becomes, in
⎡ 1 0.5 0.5⎤
general, 3 dimensional. The magnetic flux density σD ⎢
− 0.5 1 0.5⎥
B, and the electric field intensity due to the flux 12Δt ⎢ ⎥
density variation [12], ⎣⎢0.5 0.5 1 ⎦⎥
⎡ S111 S112 S 113 ⎤
⎢ 1
γ 1 ⎥
∂B ∂B =− 1
⎢ S 21 S 22 S 23⎥
∇ΧE = − (or) ∫ E.dl = − ∫ .d S (or) 2D ⎢ 1
∂t s ∂t 1 1 ⎥
J
⎢⎣ S 31 S 32 S 33⎥⎦
∂B
∫ ∇Χ E.dS = − ∫ dS ⎡1 0.5 0.5⎤
S ∂t σD ⎢ ⎥
− ⎢ 0. 5 1 0. 5 ⎥
∇ΧH = JI ; JI = JS + Je 12Δt
⎢⎣0.5 0.5 1 ⎥⎦
where Ji is the total current density. Also, let
1
γ= , we get, If Js =0 and σ = 0 , the above equation reduces to
μ
∇Χγ∇ΧA = ( JS + Je); Je = σE
[sij ][A] = 0 Generally, Js is the time dependent. An
where E is the induced electric field intensity in initial solution A(t + Δt ) =0 may be used to initial
the piece p. solution procedure. The matrix equation must be
The above equation after arranging in a format, solved iteratively to obtain the solution at every t.
A(t + Δt ) A(t ) an essential criterion for obtaining an accurate
∇.(γ∇A(t + Δt )) − σ +σ + JS (t + Δt ) = 0 solution is that Δt should sufficiently small.
Δt Δt
Applying weighted residual, B. Simulation and Results
(t + Δt)
⎡ q1q1 + r1r1 q1q2 + r1r2 q1q3 + r1r3 ⎤⎡ A1⎤ Here, a finite element method (FEM) based
γ ⎢ ⎥⎢ ⎥
= − ⎢q2q1 + r1r2 q2q2 + r2r2 q2q3 + r2r3⎥⎢A2⎥ commercial package namely ANSYS [13] is used
2D to determine stray losses in a transformer. In this
⎢⎣q3q1 + r3r1 q2q3 + r2r3 q3q3 + r3r3⎥⎦⎢⎣ A3⎥⎦
further the method of finding stray losses are
(t +Δt) (t)
⎡ 1 0.5 0.5⎤⎡ A1⎤ ⎡A1+0.5A2+0.5A3⎤ “magnetic harmonic analysis” is used. This is well
σD ⎢ σD ⎢
− 0.5 1 0.5⎥⎢A2⎥ + 0.5A1+ A2+0.5A3⎥ explained in electromagnetic field analysis guide.
12Δt ⎢ ⎥⎢ ⎥ 12Δt ⎢ ⎥
⎣0.5 0.5 1 ⎦⎣ A3⎦ ⎣0.5A1+0.5A2+ A3⎦ The following case study is solved using the same
⎡1⎤ to find the stray losses in various conducting parts
Js(t +Δt) ⎢ ⎥ of the transformer.
+ 1 =0
6 ⎢⎥
⎣1⎦
The example considered here is a concentric
Let us express the above equation in the following
wound transformer along with its stray fittings as
matrix form.
shown in fig. 1(a) below. Here shielding is
SA(t + Δt )) = Q
provided along the tank wall CRGO material .this
problem is analyzed using ANSYS to determine

239
 Journal of Theoretical and Applied Information Technology
©2005 - 2008 JATIT. All rights reserved.
www.jatit.org
the stray loss in tank walls, clampings, flitch
plates, etc[13]. The respective flux density and
flux plots are shown in fig. 1(b) and fig.1 (c).
Fig. 1(a). Transformer Model
Fig.1 (b). Flux density plot of Model shown in
fig.1
Fig.1 (c). Flux plot of Model shown in fig.1
From the above analysis the total stray loss in
various stray parts of the transformer is found and
this value is used in the hot spot and life estimation
calculations, which is illustrated in the further
sections.
240
III. THERMAL MELING OF TRANSFORMER
The winding hot spot temperature is considered to
be the most important parameter in determining
the transformer loading capability. It determines
the insulation loss of life and the potential risk of
releasing gas bubbles during a severe overload
condition. This has increased the importance of
knowing the hot spot temperature at each moment
of the transformer operation at different loading
conditions and variable ambient temperature. Thus
the appropriate thermal model to determine hot
spot temperature has become the designer’s
challenge. In this paper a novel and advanced
thermal model is proposed in view of variations in
hot spot temperature at each moment of the
transformer operation at different loading
conditions and variable ambient temperature[15].
A. Thermal model based on electrical -thermal
equivalent circuit
A thermal model of a power transformer in the
form of an equivalent circuit based on the
fundamentals of heat transfer theory has been
suggested by Swift. The proposed thermal model
was established to determine the hot spot
temperature. The top oil temperature was
calculated from the air-to-oil model. The top oil
temperature becomes the ambient temperature for
the winding to oil model. Based on this approach a
model which considers the non-linear thermal oil
resistance has been introduced by Susa. The oil
viscosity changes and loss variation with
temperature were included in the method. The
model was shown to be valid for different
transformer units. The model is analogous to the
top oil model equation. The losses used in the
model are the estimated highest losses that
generate the heat at a specific location in the LV or
HV winding. The model is easy to implement and
is validated by comparison with measured results.
In transformers, oil is typically used as the coolant.
The heat generated by losses in transformers is
taken up by the oil and carried into a heat
exchanger, which in most cases is an oil-air cooler.
The cooler dissipates heat to the surroundings by
natural or forced flow[16].
The non linear thermal resistance is related to the
many physical parameters of an actual transformer.
The exponent defining the non linearity is
traditionally n if the moving fluid is air and m if it
is oil.
 Journal of Theoretical and Applied Information Technology

©2005 - 2008 JATIT. All rights reserved.

www.jatit.org

traditionally m. The typical value used for m is 0.8.

B. Top oil thermal model The hot spot thermal equation is based on the
The top oil thermal model is based on the thermal lumped circuit shown in Fig. 3 (a) & (b).
equivalent thermal circuit shown in fig.2 (a) and
(b). A Simple RC circuit is employed to predict the
top oil temperature.

In the thermal model all transformer losses are

represented by a current source injecting heat into
the system. The capacitances are combined as one
lumped capacitance. The thermal resistance is
represented by a non-linear term. The differential
equation for the equivalent circuit is

where,
qTot is the heat generated by total
losses, W
Cth-oil is the oil thermal capacitance, Wmin
/0 C
Rth-oil is the thermal resistance Fig. 3 (a) & (b) Thermal Model for hot spot
Θoil is the top oil tempetature, 0C temperature

The differential equation for the equivalent circuit

is

where,
qW is the heat generated by the losses at hot spot
location, W
Cth-H is the winding thermal capacitance at the hot
spot location
Rth-oil is the thermal resistance at the hot spot
location, 0C/ W
Θoil is the top oil temperature at the hot spot
location, 0C

The above equation is reduced to

Fig. 2 (a) & (b) Thermal Model for top oil

temperature

C. Winding hot spot thermal model

In the thermal model the calculated winding losses
generate the heat at the hot spot location. The
thermal resistance of the insulation and the oil
moving layer is represented by a non-linear term. τHis the winding time constant at hot spot
The exponent defining the non linearity is location, min.

241
 Journal of Theoretical and Applied Information Technology

©2005 - 2008 JATIT. All rights reserved.

www.jatit.org

The variation of losses with temperature is

included in the equation above using the resistance
correction factor.

IV. SIMULATION RESULTS

The transformer input parameters needed for the
thermal model is shown in Table 1.

Table 1: Input parameters of thermal model

Fig. 5 (a) Thermal model
Rated top oil rise over ambient 38.30 C
Rated hot spot rise over top oil 20.30 C
Ratio of load losses to no load losses 6.200 C
pu eddy current losses 0.65
MW
at hot spot location, LV
pu eddy current losses 0.3
at hot spot location, V
Top oil time constant 170 min
Hot spot time constant 6 min
Exponent n 0.9
Time (h)
Exponent m 0.8
Fig. 5(b) Input load cycle for thermal model
The following fig.4 (a) and (b) represents the top
oil temperature model of 250 MVA Transformer
and its temperature graph.
0
C

Time (h)
Fig. 5 (c) Hot spot temperature

From the above plot it is clear that for a given load

cycle the hot spot temperature rises from rated
Fig. 4(a) Top oil equation model temperature to a high enormous value.

Time (h)
Fig. 4(b) Calculated top oil temperature graph
The following fig. 5(a) and (c) shows the
Thermal model of 250 MVA Power Transformer
and hot spot plot for the in put load cycle of
current shown in fig. 5 (b).
V. CONCLUSION
The thermal models are fully dependent on
accurately defined steady-state temperature rises.
Therefore, it will be important to develop a steady-
state calculation method, especially because most
installed transformers are not heat-run tested.
However, further research and development is
needed to improve the existing monitoring systems
and introduce new solutions that include the
transformer thermal models and their real-time
application. Utilizing thermal models allows both
transformer manufactures and users to run
different loading and ambient scenarios and, by
analyzing the results, improve the transformer
design (costs size and load carrying capacity)[17-

242
 Journal of Theoretical and Applied Information Technology
©2005 - 2008 JATIT. All rights reserved.
www.jatit.org
18]. An electromagnetic analysis using a finite
element model has been adapted to predict
transformer winding losses. This can be used to
calculate the eddy losses in individual turns/discs
to enable location of the winding losses that cause
the hot spot and to predict the hot spot factor H.
In this paper a novel thermal model has
been established to determine the hot spot
temperature. The top oil temperature is calculated
from the top oil equation. The top oil temperature
becomes the ambient temperature for a hot spot
equation model. The equations are modeled in
Simulink and validated using transformer data
from measurements in the factory, thus a real time
online monitoring system can be developed and
used.
REFERENCES
[1] Working Group 12-09, “Analytical
Determination of Transformer Windings Hot-
Spot Factor” Electra, no.161, Aug 1995, pp.29-
33.
[2] D. Pavlik, D. C. Johnson, and R. S. Girgis,
“Calculation and reduction of stray And eddy
losses in core form transformers using a highly
accurate finite element Modeling technique,”
IEEE Trans Power Delivery, vol. 8, no. 1 Jan.
1993, Pp.239-245.
[3] R. Komulainen and H. Nordman, “Loss
evaluation and use of magnetic and
Electromagnetic shields in transformers,” Paper
no. 12-03, CIGRE 1988.
[4] IEC 354 Standard Publications: 1991, Loading
Guide for Oil-Immersed Power Transformers.
[5] IEEE standard C57.91-1995, Loading Guide
for Mineral-Oil-Immersed Immersed
Transformers.
[6] Working Group 12-09, “Heat Run Test for
Power Transformers” Electra,
no.129, pp.37-39.
[7] Working Group 12-09, “Experimental
Determination of Power Transformer Hot- Spot
Factor” Electra, no.161, Aug 1995, pp.35-39.
[8] IEC Publication 76-2:1993, Power
Transformers, Part 2. Temperature rises.
243
[9] (Alegi, 1990) Alegi, G.L., and Black W.Z.,
"Real-time thermal model for an oil immersed,
forced-air cooled transformer", IEEE
Transactions on Power Delivery, Vol. 5, Iss. 2,
April 1990, pp.991-999.
[10]A. Konard, “Inegrodifferential Finite Element
Formulation of Two- Dimensional Steady-
State Skin Effect Problems,” IEEE Trans. on
Magnetics, vol. MAG-18, no.1, Jan 1982,
pp.284-292.
[11]J. Weiss and Z. J. Csendes “A One Step Finite
Element Method for Multiconductor Skin
Effect Problems,” IEEE Trans. on Power
Apparatus and Systems, vol. PAS-101, no.10,
Oct. 1982, pp.3796-3800.
[12]S.V.Kulakrani, S.A.Khaparde
“TRANSFORMER ENGINEERING DESIGN
PRACTICES”.
[13]Finite element method in electrical power
engineering A.B.J.Reece, T.W...Presto.
[14]Electromagnetic field analysis guide, ANSYS.
[15] Working Group 12-09, “Experimental
Determination of Power Transformer Hot-
Spot Factor” Electra, no.161, Aug 1995,
pp.35-39.
[16] (Alegi, 1990) Alegi, G.L., and Black W.Z.,
"Real-time thermal model for an oil
immersed, forced-air cooled transformer",
IEEE Transactions on Power Delivery, Vol. 5,
Iss. 2, April 1990, pp.991-999.
[17] (Aubin, 1992) J. Aubin, and Y. Langhame,
"Effect of oil viscosity on transformer
Loading capability at low ambient
temperatures", IEEE Transactions on Power
Delivery, Vol.7, Iss.2, April 1992, pp.516-
524.
[18] (Betta, 2000) Betta, G., Pietrosanto A., and
Scaglione A., "An enhanced fiber optic
temperature sensor system for power
transformer monitoring", Instrumentation and
Measurement Technology Conference, 2000.
IMTC 2000. Proceedings of the 17th IEEE,
Vol.1, 1-4 May 2000, pp.153-158.